Re: scaling up is the answer?
"Eventually, maybe, we'll have a useful reactor that's smaller than, say, a skyscraper."
Skyscraper sized - if we stick with Tokomak, yes, could be.
But of course, they SHOULD do it. At least one. As a proof of concept.
Like the first 'general use' computer, which was ginormous and had to be programmed by re-wiring it. That wouldn't be the specialized 'Collossus' machine developed by Turing and the Bletchley Park group [I don't know how that one was programmed, exactly] which was not 'general purpose', but still pretty damn good at the time. I think the first 'general use' computer was Eniac, which had to be re-wired to re-program it. And it was ginormous. That's the point.
As for my preference, there's a design that's based on keeping the plasma flow all at the same velocity, that twists and turns in somewhat unusual ways, producing a 'ribbon' effect as I recall. I can't remember what it is called, though...
And there are other possibilities with 'linear' rather than 'cyclic' accelerators.
And I don't know if they're at least looking into RESONANCE ENERGIES and other physical effects to assist with the fusion process. A linear system could employ something similar to a 'travelling wave tube' to bunch the protons/deuterons together at ideal energy levels, as one example. It would be a type of 'resonance confinement'.
Anyway you'd think they'd be talking about this stuff if it were being done, right?